Maintaining consistent traceability in high precision isotope measurements of CO<sub>2</sub>: verifying atmospheric trends of δ<sup>13</sup>C

(1)

AMTD

5, 4003–4040, 2012

Maintaining traceability

L. Huang et al.

Title Page

Abstract Introduction

Conclusions References

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Printer-friendly Version

Interactive Discussion

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Atmos. Meas. Tech. Discuss., 5, 4003–4040, 2012 www.atmos-meas-tech-discuss.net/5/4003/2012/ doi:10.5194/amtd-5-4003-2012

Atmospheric Measurement Techniques Discussions

This discussion paper is/has been under review for the journal Atmospheric Measurement Techniques (AMT). Please refer to the corresponding final paper in AMT if available.

Maintaining consistent traceability in high

precision isotope measurements of CO

₂

:

verifying atmospheric trends of

δ

13

_C

L. Huang, A. Chivulescu, D. Ernst, W. Zhang, and Y.-S. Lee

Climate Research Division (CRD), Atmospheric Science & Technology Directorate (ASTD), Science and Technology Branch (STB), Environment Canada (EC) 4905 Dufferin Street, Toronto, Ontario M3H 5T4, Canada

Received: 12 March 2012 – Accepted: 24 April 2012 – Published: 6 June 2012 Correspondence to: L. Huang (lin.huang@ec.gc.ca)

(2)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Abstract

Maintaining consistent traceability of high precision measurements of CO2 isotopes is critical in being able to observe accurate atmospheric trends ofδ13C (CO2). Although a number of laboratories/organizations around the world have been involved in baseline measurements of atmospheric CO₂isotopes for several decades, the reports on their

5

traceability measures are rare. In this paper, a principle and an approach for the trace-ability maintenance of high precision isotope measurements (δ13C and δ18O) in at-mospheric CO₂is described. The uncertainties of the traceability have been estimated based on the history of annual calibrations over the last 10 yr. The overall uncertain-ties of CO2 isotope measurements for individual ambient samples carried out by our

10

program at Environment Canada are estimated (excluding the uncertainty associated with the sampling). The values are 0.02 ‰ and 0.05 ‰ inδ13C andδ18O, respectively, close to the WMO targets for data compatibility. The annual rate of change in δ13C of the primary anchor used in our program (which is the laboratory standard linking ambient measurements back to the primary VPDB scale) is close to zero (−0.0016±

15

0.0012 ‰ per year) over the period of 10 yr (2001–2011). The average annual decreas-ing rate ofδ13C in air CO₂ measurements at Alert over the period from 1999 to 2010 has been confirmed and verified, which is−0.025±0.003 ‰ per year. The total change

of δ13C in the annual mean value during this period is ∼ −0.27 ‰. The concept of

“Big Delta” is introduced and its role in maintaining traceability of the isotope

measure-20

ments is described and discussed extensively. Finally, the challenges and a strategy for maintaining traceability are also discussed and suggested.

1 Introduction

Precise determination of the isotope compositions of atmospheric CO2 plays an im-portant role in understanding carbon cycle, in turn, addressing the issue of the

con-25

tinuous increase of atmospheric CO₂at regional and global scales. Numerous studies

(3)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

have been conducted to understand the exchanges of CO₂ between the atmosphere, the terrestrial biosphere and the oceans to quantify the relevant sources and sinks (Keeling 1960, 1961; Keeling et al., 1979, 1995; Mook et al., 1983; Francey et al., 1995; Bakwin et al., 1998; Ciais et al., 1995, 1997; Battle et al., 2000; Allison and Francey, 2007). From 1998 through 2010, an average annual global rate of change

5

of carbon isotopic composition in atmospheric CO2 is ∼ −0.026±0.001 ‰ in δ 13

C per year. The result is derived from annual averages of all surface mbl references. The mbl references are constructed (Masarie and Tans, 1995) using NOAA surface flask data from the Global Cooperative Air Sampling Network (http://www.esrl.noaa. gov/gmd/ccgg/about/global means.html). The CO₂isotope data from the network were

10

analyzed by The Institute of Arctic and Alpine Research (INSTAAR) at University of Colorado (White and Vaughn, 2011). On a regional scale the corresponding value is very similar e.g.∼ −0.025±0.003 ‰ yr−1 from 1999 through 2010, obtained by

Envi-ronment Canada, at Alert GAW station (Fig. 1a, b). The magnitudes of those changes, together with the corresponding CO2concentration changes, were driven by the

con-15

tributions from various carbon sources and sinks (including natural and anthropogenic ones). Precisely identifying the magnitude of those changes will help us to understand the complicated mechanisms of carbon cycle and track the human-induced CO2 in the atmosphere. This task is especially challenging as the gradients as well as the trends in atmosphericδ13C are very small compared to the levels of analytical

preci-20

sion even by the most accurate measurement techniques applied (e.g. Isotope Ratio Mass Spectrometer, IRMS). That’s why the World Meteorological Organization/Global Atmospheric Watch (WMO/GAW) measurement community has strongly encouraged making high precision measurements in the order of 0.01 ‰ and 0.05 ‰ forδ13C and

δ18O, respectively (Expert Group recommendations in GAW publications 161, 168,

25

(4)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

changes of isotope standards are much less than 0.02 ‰. Isotope standards play an important role in maintaining consistent traceability (Huang et al., 2002). A consistent traceability is essential for the observed data being used to derive the trend. To imple-ment a traceability for isotope measureimple-ments in flask air CO2samples, various forms of lab-standards are used, such as pure CO₂ in high pressure cylinders, air CO₂ and

5

CO2 produced from pure carbonates (Huang et al., 2002; Mukai et al., 2005; Allison and Francey, 2007; Brand, et al., 2009). Usually, more than one level (e.g. primary, secondary, etc.) and more than one form of standards are used in individual labora-tories (Huang et al., 2002; Allison and Francey, 2007). It is very challenging, however, to ensure that the uncertainties of those standards are less than 0.02 ‰ inδ13C (i.e.,

10

the standards are stable within the range of ∼0.02 ‰) over a long time period. The

time span should be comparable to this from which the atmospheric trend is derived. This is difficult because ensuring that a lower level standard is stable (or not drifting) within the range of∼0.02 ‰ requires a higher level standard with a better stability (i.e.,

<0.02 ‰). Usually, the higher level standard is used for the calibration of the lower level

15

standard. Ultimately, it is imperative to ensure that the primary standard (i.e. NBS19) is stable over decadal time periods with an annual change rate significantly less than 0.02 ‰. The crucial questions arise therefore: Is this primary standard, NBS19 abso-lutely stable? How do we know? How well we know? And if not, how could we assess the uncertainty and stability of a standard? What is its impact on the uncertainties of

20

CO2isotope measurements for individual ambient samples?

In this paper, we present our results obtained over the last 10 yr to illustrate how these questions are addressed, including:

– the traceability of high precision CO2isotope measurements for our program;

– the entire records in annual calibrations of secondary standards directly against

25

NBS19-CO2) over the period of 10 yr, which demonstrate the traceability imple-mentation and maintenance in our program;

(5)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

– the stability of the primary anchor (a secondary standard) which links the individ-ual measurements back to the primary VPDB-CO₂scale and implies the stability of NBS19. This can be used for verifying a long-term trend ofδ13C in atmospheric CO2at Alert;

– the uncertainty of the traceability (which is the overall uncertainty propagated from

5

all different levels of standards used in the traceability chain, including NBS19-CO₂production from carbonates) and the discussion;

– the challenges and recommendations based on the results, which can be used as a strategy to tackle these problems.

2 Traceability

10

Metrological traceability is defined as “property of a measurement result whereby the result can be related to a reference through a documented unbroken chain of calibra-tions, each contributing to the measurement uncertainty” (GAW Report No. 194). No matter what kind of and how many level of standards are used for calibrations, CO2 isotope measurements must be finally traced back to the primary scale (VPDB) via the

15

primary standard NBS19-CO2. It is known that VPDB is a hypothetical standard. As the supply of PDB is exhausted, the primary VPDB scale is established by adopting the isotopic compositions of NBS19 relative to VPDB as:+1.95 ‰ forδ13C_NBS19_/_VPDB and−2.2 ‰ forδ18ONBS19/VPDB. However, NBS19-CO2 only defines one point on the primary scale or ruler. It is almost impossible to accurately calibrate other secondary

20

standards by the one point ruler with no unit defined. In order to define the unit on the primary ruler, at least two standards are required (assuming instrument liberality). It’s better to have three standards, thus the linearity of the instrument can be assessed. Following this principle, establishing a secondary scale (i.e., a local scale for individ-ual program), at least two standards (three are better) are needed and a large isotope

25

(6)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

used in the traceability chain to link the individual flask-air CO₂isotope measurements back to the primary standard. One is the primary (i.e., NBS19) and the other is the sec-ondary (i.e., lab carbonate standards), including NBS18, Cal1 and Cal2. NBS19 and NBS18 are international reference materials, which were purchased from the Inter-national Atomic Energy Agency (IAEA) through the web (http://nucleus.iaea.org/rpst/

5

ReferenceProducts/ReferenceMaterials/Stable Isotopes/13C18and7Li/index.htm; last access: 26 May 2012). NBA19 was obtained from a limestone (mainly composed of CaCO3) with an unknown source, whereas NBS18 is a calcite (CaCO3) from Fen, Nor-way. NBS19 and NBS18 are used to define the unit on the primary scale. Both Cal1 and Cal2 are calcium carbonates purchased from Aldrich Chemicals and Fisher Scientific,

10

respectively. Cal1 and Cal2 are used to build the link between the individual measure-ments and the primary standard and at the same time to ensure the stability of the primary anchor (discuss later). As shown the schematic of the traceability pathway in Fig. 2 the implementation of the traceability in our program includes two operational steps, i.e. annual calibration and routine measurement.

15

During the annual calibration periods, the secondary standards are calibrated by the primary standard, i.e. NBS19-CO₂. Each individual calibration period was conducted via measuring those secondary standards together with NBS19-CO2against the same Working Reference Gas (WRG) within one day time span (according to “the identical treatment” principle). The isotopic compositions of these standards are traced to the

20

primary standard by the following equations.

RLab-Std/RVPDB-CO2=[RLab-Std/RWRG]×[1/(RNBS19-CO2/RWRG)]×(RNBS19-CO2/RVPDB-CO2)

=[R_Lab-Std/R_NBS19-CO2]×(RNBS19-CO2/RVPDB-CO2)

=[∆Lab-Std/NBS19-CO2×10−3+1]×[∆NBS19-CO2/VPDB-CO2×10−3+1] (1)

25

whereR: the ratio of [mass 45/mass 44] or [mass 46/mass 44] in CO₂

∆45 or 46_{Lab-Std/NBS19-CO2}=[(RLab-Std−RNBS19-CO2)/RNBS19-CO2]×103‰

(7)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

This term∆45 or 46_A_/_B is introduced as Big Delta. It is the relative deviation of isotopic ratio (usually given in ‰) between two materials. It can also be expressed as [(RA/RB)−1]× 103‰. For our case, A is the Lab-Std and B is NBS19-CO2. Although the expression of∆45 or 46_A_/_B appears identical as the definition of the small delta between A and B (i.e.,

δ_A45 or 46_/_B =[(RA/RB)−1]×10 3

) ‰, the determination of Big Delta should not be obtained

5

by a direct measurement against each other because the identical treatment principle should be applied to both A and B. A Big Delta value should be derived from the two raw measurements which are conducted separately against the same WRG, as shown in Fig. 2.

By definition, a Big Delta value is independent of the WRG. It is, however,

depen-10

dent on the configuration and the degree of cleanliness of the ion source in an IRMS, which determines the magnitude of possible cross-contaminations in the source (Meijer et al., 2000). Any physical or configuration modification to the ion-source or changes in electronics (amplifiers) can either increase or decrease the Big Delta values, due to decreases or increases the magnitude of cross-contamination (Meijer et al, 2000).

15

Under ideal clean conditions of an IRMS, a Big Delta value is close to a constant, and thus can be precisely determined. Using Big Delta approach, the units on the primary scale can be defined.

The first two terms on the right side of Eq. (1) are measured during the annual calibration phase and the second term is the constant recommended by International

20

Atomic Energy Agency (Craig, 1957; Allison et al. 1995). According to Eq. (1), the an-nual calibrations is to determine the Big Delta values between the secondary laboratory standards and NBS19-CO2(i.e., [∆Lab-Std/NBS19-CO2]), which can be also considered as determining the units on the primary scale. During each individual calibration, three sets of pure CO₂ampoules are prepared from acid digestions of carbonates as shown

25

in the following reaction:

(8)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Each set includes NBS19-CO2 (+1.95 ‰ in δ 13

CVPDB-CO2 and −2.2 ‰ in

δ18O_VPDB-CO2), NBS18-CO₂(∼ −5 ‰ inδ13CVPDB-CO2 and∼ −23 ‰ inδ 18

O_VPDB-CO2), Cal1-CO2 (∼ −46 ‰ in δ

13

CVPDB-CO2 and ∼ −22 ‰ in δ 18

OVPDB-CO2) and Cal2-CO2 (∼ −2.6 ‰ in δ13CVPDB-CO2 and ∼ −12.6 ‰ in δ

18

OVPDB-CO2). The corresponding Big Delta values (i.e., _∆_{NBS18-CO2/NBS19-CO2}, _∆_Cal1_/_NBS19-CO2, _∆_Cal2_/_NBS19-CO2, _∆_Cal2_/_Cal1

5

and etc.) are determined. The measurement protocol is shown in Table A1. The Big Delta values of ∆NBS18-CO2/NBS19-CO2 are used as the criteria for the quality assur-ance/quality control (QAQC) of the Big Delta values of∆Cal1/NBS19-CO2,∆Cal2/NBS19-CO2, and∆Cal2/Cal1since the measurements of NBS18 can be compared with the literature values. The details for obtaining valid Big Delta values during annual calibrations are

10

described in Fig. 2.

During routine measurements, individual flask CO2 samples are analyzed in con-junction with the secondary standards against the same WRG. Usually, 12 samples are measured as a suite during a single day period. A pair of Cal1 and Cal2 are an-alyzed before and a Cal2 is anan-alyzed after the suite. The measurement protocol is

15

described in Table A2. The derivation of Big Delta is included in every single daily mea-surement suite. The comparison of this value with the annual determined Big Delta value provides an important validation measure of the secondary standard production. The second Cal2 analyses at the end of the daily measurement suite provides an addi-tional measure of system stability over the entire measurement period and an important

20

validation measure for the unknown samples in the suite. The isotopic composition of individual samples are determined using the following equation and traced back to the primary standard.

(9)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

[RSam/RVPDB-CO2]=[RSam/RWRG]×[1/(RLab-Std/RWRG)]×(RLab-Std/RVPDB-CO2)

=[RSam/RLab-Std]×(RLab-Std/RVPDB-CO2)

=[RSam/RLab-Std]×[RLab-Std/RNBS19-CO2]×(RNBS19-CO2/RVPDB-CO2)

=[_∆_Sam_/_Lab-Std×10−3+1]×[∆Lab-Std/NBS19-CO2×10−3+1]

×[∆NBS19-CO2/VPDB-CO2×10−3+1] (3)

5

where RSam: the ratio of mass 45 to 44 or mass 46 to 44 in a sample.

δ45(CO2)Sam-VPDB-CO2andδ 46

(CO2)Sam-VPDB-CO2are defined as:

δ45/or 46_(CO

2)Sam/VPDB-CO2=[(RSam/RVPDB-CO2)−1]×103‰ (4)

10

Equation (3) shows the documented traceability chain in CO₂ isotope measurements for individual CO2samples collected in Environment Canada Greenhouse Gas Obser-vation Network (Huang and Worthy, 2005) back to the primary scale. Here VPDB-CO2 represents the CO₂ evolved from VPDB (Vienna PeeDee Belemnite). Using the re-sults of Eq. (4), δ13C_Sam_/_VPDB-CO2 and δ18O_Sam_/_VPDB-CO2 are calculated by applying

15

the17O correction used by Allison et al. (1995). This correction is very similar to the Crag correction (Craig, 1957). It should be noted that air samples have an additional N2O correction requirement as N2O is an interference (also with masses 44, 45, 46) for CO2isotopic ratio measurements. The same correction algorithms for both

17 O and N₂O are applied to the entire dataset.

20

Based on Eq. (3), routine measurements is to determine the Big Delta value be-tween the sample and the lab standards. Lab standards are used to anchor the un-known individual measurements to the primary scale (VPDB-CO₂). In general, a lab-oratory standard calibrated directly or indirectly using NBS19-CO2 and used to de-termine δ13CSam/VPDB-CO2 and δ

18

OSam/VPDB-CO2 is referred to as a primary anchor.

25

(10)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

from Eq. (3) via error propagations of the two terms on the right (i.e,∆Sam/Lab-Std and

∆Lab-Std/NBS19-CO2). One of the uncertainties is related to the primary anchor. Advan-tages of using this approach include:

i. the units of the primary scale are evaluated every year;

ii. individual isotopic measurements are firmly anchored to the primary scale;

5

iii. the uncertainty of the primary anchor can be characterized, monitored and mini-mized;

iv. and the overall uncertainty of individual ambient measurements can be explicitly estimated.

3 Uncertainty in traceability

10

To maintain a consistent traceability for isotope measurements in our program, two secondary standards (i.e., Cal1 and Cal2) are used with a significant isotopic diff er-ence (∼42 ‰). They, along with NBS18, are directly calibrated by NBS19 on an annual

basis. During the calibration periods (usually between February to April when the rel-ative humidity is typically the lowest of a year and when the most stableδ18O values

15

can be attainable). NBS19, NBS18, Cal1 and Cal2 are evolved into pure CO₂ via acid digestions using the H3PO4(with mass percentage>100 % and the specific gravity of 1.91–1.92 at 25±0.1◦C). It is known that the amount of H2O in H3PO4 impacts the precision of δ18O analysis because the oxygen isotopes in evolved CO2 can easily exchange with those in liquid H2O (e.g. McCrea 1950; Clayton, 1958). Equation (2)

20

indicates that that some water will be released from the reaction together with the CO₂. To minimize the impact of available liquid H2O on the isotopic exchanges between CO2 and H2O while the CO2 is being evolved from the reaction, excess P2O5 is needed to absorb the H2O which can potentially exist in the acid (Sharp, 2007). A solution of H₃PO₄ with a mass percentage of greater than 100 % indicates some excess P₂O₅ in

25

(11)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

the solution. If the mass percentage is too large, the solution tends to crystallize and makes the di_ffusion of CO₂to the gas phase more di_fficult (Sharp, 2007). This will also cause isotopic fractionations and affect the precisions of the δ18O values. Commer-cially available H3PO4 has mass percentages on the order of only 85 % and thus not suitable. H₃PO₄with a mass percentage greater than 100 % can be only custom-made.

5

The specific gravity of 100 % H3PO4is approximately 1.86 g ml− 1

. Based on our expe-riences, the ideal range of the specific gravity of H3PO4is from 1.91 to 1.92 g ml−

1 . Our in-house procedure of making the H₃PO₄ (>100 %) is attached as Appendix 2. A re-cent report by Wendeberg et al. (2011), found that theδ18O of H3PO4 will likely affect theδ18O of CO₂evolved from the acid digestion when the mass percentage of H₃PO₄

10

is<102 %. This is likely due to isotopic exchanges between H2O and H3PO4and be-tween H2O and CO2. Since it’s important to use>100 % H3PO4, the specific gravity of H₃PO₄used in our program over the past 10 yr ranges from 1.91 to 1.92 g ml−1

(corre-sponding to a mass percentage of 104 %–105 %). During individual calibration periods for each of the four carbonates, there are, at least, three separate acid digestion are

15

processed, followed by cryogenic extractions (of the evolved CO₂) and IRMS measure-ments. The Big Delta values, i.e., [∆Lab-Std/NBS19-CO2] between the secondary labora-tory standards and NBS19 are obtained.δ13CLab-Std/VPDB-CO2andδ

18

OLab-Std/VPDB-CO2 are determined, as described in Sect. 2. These data obtained from the calibration pe-riods over the past decade (Tables 1–4) are sufficient for estimating the uncertainties

20

and the stability of the primary anchor. In turns, theses results are used to assess the overall uncertainty of individual measurements and the stability of the traceability. To do this, the uncertainties of the two terms on the left side of Eq. (3) need to be esti-mated. The annual calibration data over last ten years (2001–2010) by using two dif-ferent IRMSs (Finnegan MAT252 and Micromass, IsoPrime) are shown in Tables 1–4.

25

(12)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

same as these forδ18O). One can see, that the standard deviation is proportional to the magnitude of absolute Big Delta value (Fig. 3). The smaller the absolute Big Delta value, the smaller its standard deviation, indicating that the cross contamination e_ffect in the ion source has a larger impact on these Big Deltas with bigger absolute values than these Big Delta values with the smaller absolute values. The closer the absolute

5

Big Delta value to zero, the smaller the cross contamination e_ffect and thus obtaining the least possible standard deviations (Fig. 3). To minimize the cross contamination effect on both∆Sam/Lab-Std and∆Lab-Std/NBS19-CO2, it is suggested to use a lab standard with itsδ13C value between the NBS19 (i.e.,₊1.95 ‰) and the ambient atmospheric CO2 (∼ −8 ‰) as the primary anchor. As mentioned, Cal2 is used as the primary

an-10

chor for all flask samples collected from Environment Canada Greenhouse Gas(GHG) sampling network because t has the smallest absolute Big Delta values with NBS19 (i.e.,∆45Cal2/NBS19 and ∆

46

Cal2/NBS19) as well as the smallest standard deviations in both

∆45Cal2/NBS19 and ∆ 46

Cal2/NBS19. As shown in Table 3, the measured 1 sigma uncertainty of_∆45_{Cal2/NBS19-CO2} and_∆46_Cal2

/NBS19-CO2 is<0.02 ‰ and<0.04 ‰, respectively. It is also

15

shown that Cal2 during the 10 yr period shows the greatest stability amongst all of the secondary standards (Fig. 1c), including NBS18. The uncertainty of the second term (∆Sam/Lab-Std) can only be determined by using an air-CO2standard due to the need for replicate analysis over many years. Flasks are not suitable due to the limitation of hav-ing only shav-ingle analyses with our measurement procedure. A high pressure air-CO₂

20

tank, which is used for the purpose of quality control (QC), is treated in exactly the same manner as air flasks. The uncertainties of∆QC air-CO2 tank/Lab-Stdis determined by repeated analysis over several years. The QC air-CO₂ tank used here for deriving the uncertainty of∆Sam/Lab-Std is designated as QC3, and was filled at Alert, GAW station. As shown in Fig. 4, the measured 1 sigma uncertainty for∆45_QC3_/_Cal2 and ∆46_QC3_/_Cal2is

25

0.017 ‰ and 0.043 ‰, respectively. After applying the17O and N2O corrections, the magnitude of the uncertainty in δ13C and δ18O are the same as those in ∆45_QC3_/_Cal2

(13)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

and∆46_QC3_/_Cal2, respectively, inferring that the uncertainties are primarily caused by the analytical procedures (i.e., acid digestions and IRMS measurements). Now that the un-certainties for∆Lab-Std/NBS19-CO2(i.e.,∆Cal2/NBS19-CO2) and∆Sam/Lab-Std(i.e.,∆QC3/Cal2) values have attained, the overall uncertainties in Eqs. (3) and (4) can be determined.

According to the principle of propagation of independent errors, the uncertainty of

5

Eq. (3) can be calculated using Eq. (C2) in Appendix C, i.e.,

σ[δ45/46(CO₂)_QC3_/_VPDB-CO2]=[(σ∆_QC345/46_/_Cal2)2+(σ∆_Cal245/46_/_NBS19-CO2)2]1/2

Using the results from Fig. 4 for σ (_∆45/46

QC3/Cal2) and Table 3 for σ (∆ 45/46

Cal2/NBS19-CO2), the propagated uncertainty forδ45RQC3/VPDB-CO2andδ

46

RQC3/VPDB-CO2 is 0.02 ‰ and

10

0.05 ‰, respectively. These are very close to the values shown in Fig. 4 which were based on statistical calculations of the individual measurements forδ13C andδ18O. It is safe to conclude, that the overall uncertainty of the traceability in CO2isotope mea-surements for individual flask samples from our network is on the order 0.02 ‰ forδ13C and 0.05 ‰ forδ18O. No uncertainties in the17O and N₂O corrections are considered

15

since the same algorithm has been applied since the inception of the program.

4 Verifying a long-term trend ofδ13_{C at alert}

Determining accurate long-term trends and mean annual change rates ofδ13C in atmo-spheric CO2is critical for understanding source and sink changes with time and as well as for assessing the effectiveness of carbon emission-control policies and mitigation

20

measures. A long-term atmospheric trend inδ13C can only be determined if the stabil-ity of the primary anchor on VPDB-CO2is known. As noted above, Cal2 has shown to be the most stable standard amongst the carbonate secondary standards. As shown in panel c of Fig. 1., the annual rate of change of the primary anchor (Cal2) inδ13C over the period (from 2001 to 2011) was essentially zero (i.e.,−0.0016 ‰). This is much

(14)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

less than the analytical uncertainties of the IRMS (∼0.01 ‰). As shown in Fig. 4, the

annual drift rate ofδ13C andδ18O for the QC3 air standard is+0.003 ‰ and−0.007 ‰,

respectively. Based on these results, it can be verified that the average annual change rate of the measuredδ13C in air CO₂at the Alert station is−0.025±0.003 ‰ per year.

This likely reflects real changes in relative contributions of sources and sinks to the

5

atmosphere CO2, not due to drifts in the instrumentation and the isotope standards or analytical procedures applied. The total 10 yr change of the observedδ13C in annual average value is∼ −0.27 ‰, from−8.22 ‰ in 1999 to−8.49 ‰ in 2010 (Fig. 1).

5 The roles of Big Delta

It is known that a measured isotopic ratio (by IRMS) varies with time as well as with

10

di_fferent instruments. The latter is mainly due to di_fferent instrument design and op-erational conditions such as, physical configurations, material used, voltage applied, temperature, pressure or vacuum etc. To the best of our knowledge, current technolo-gies and instruments are not able to measure the absolute isotope ratio of a sample but only the ratio of the isotopic ratio of the sample relative to another sample. If two

15

samples have intrinsic and distinguishable isotopic compositions, the relative deviation in isotope ratio (in ‰) should be constant and independent on the fluctuations of in-strument response, as illustrated in Fig. 5. A relative deviation of two isotope ratios, i.e., a Big Delta value, in principle, can be precisely measured on an IRMS. This al-lows Big Delta to play various roles in high precision isotopic measurements. Over the

20

last decade, Big Delta has played two major roles in maintaining traceability of isotope measurements in our program.

The first role is to link all individual measurements to the primary scale (as shown by Eq. 3). As described in Sect. 2, the annual calibration and routine measure-ments are two independent steps in the chain of the traceability for our CO2 isotope

25

measurements, expressed as two Big Delta values. The uncertainties of the two Big Delta terms determine the overall uncertainty of the traceability pathway. Over the 10 yr

(15)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

period, different batches of NBS19 and NBS18 were used as well as a new ion source that was installed in 2007. In addition, the original voltage to frequency conversion (VFC) resistor was replaced in 2008 (noted in Tables 1–4). These changes can poten-tially modify the Big Delta values from the intrinsic values and contribute to their varia-tions and uncertainties. However, as shown in Tables 1–4, Big Delta values are close

5

to constant over the 10 yr period although theδ45/or 46 values vary with time due to the use of various Working Reference Gases. This indicates that, generally, the procedures for acid digestion and the instruments for IRMS analysis have been consistent over the entire 10 yr period. A closer investigation at the Big Delta values (by MAT252) shows that the Big Delta values have also slightly varied or shifted, particularly for∆Cal2/Cal1

10

values (i.e., the largest and the most sensitive to procedure/or instrument fluctuations). Based on Table 4, it is likely that the cleanliness of the ion source of the MAT252 (indi-cated by “BG count”, i.e., background count) has been a dominant factor for the small drifting over the period, particularly for∆45. As the ion source gets cleaner, the corre-sponding Big Delta values get larger. The largest Big Delta values are the bench marks,

15

which are the closest to the intrinsic/or true values. Usually, the bench mark values are obtained under the cleanest condition of an ion source. Nevertheless, it is shown that the variations of∆Cal2/NBS19-CO2by MAT252 are very small and its standard deviations (including all errors from the analytical procedures) are within the ranges of<0.02 ‰ in

δ13C and∼0.04 ‰ inδ18O over this 10 yr period. These values are close to the WMO

20

targets of 0.01 ‰ and 0.05 ‰ for data comparability inδ13C andδ18O, demonstrating a consistency in the precision and the stability of Cal2 as the primary anchor. The results on IsoPrime show very similar patterns for∆45_Cal2_/_NBS19-CO2 but not for∆46_Cal2_/_NBS19-CO2. This suggests that the di_fference in the high vacuum and the water-content levels in-side of ion source may have larger impacts onδ18O thanδ13C measurements. These

25

results indicates that as long as the magnitude of a Big Delta value is relatively small (∼10 ‰ or less as shown in Fig. 3), even the fluctuations in cleanliness of the ion source

(16)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

to conclude that Cal2, when used as the primary anchor, has precisely linked all of our isotopic measurements of atmospheric CO2samples to the VPDB-CO2scale.

The second role is to carry out quality assurance and quality control (QAQC), as a diagnostic tool, in monitoring fluctuations of instruments and associated apparatuses. It also serves as a measure to track the stability of various levels of standards. It can

de-5

tect scale drifting in time within one individual laboratory or scale contractions between laboratories. A two-point-scale-normalization in carbon isotope measurements is rec-ommended by Coplen et al. (2006) in order to resolve the issues of scale contraction or shifting via normalizing the difference betweenδ13CNBS19/VPDB and δ

13

CL-SVEC/VPDB obtained from individual laboratories to a fixed value (i.e., the recommended value).

L-10

SVEC is a lithium carbonate prepared by H. Svec, Iowa State University, originally to be used as a reference material for lithium isotopic composition (Flesch et al., 1973). Due to its quite negativeδ13C value (−46.6±0.2 ‰), it is recommended to use NBS19 and

L-SVEC together to implement a two points calibration. However, to implement this recommendation, the Big Delta values (i.e.,∆45_L-SVEC/46 _/_NBS19-CO2) between NBS19 and

15

L-SVEC in individual labs should be determined annually to track the consistency of the normalization over time. The instrument and procedure fluctuations are major error sources causing the uncertainty in traceability, leading to non-consistency of the mea-surements over time. It is reasonably assumed that changes in standard or references themselves are also possible. This in turn can also cause changes in Big Delta values.

20

However, distinguishing the fluctuations due to external factors from the changes due to standards themselves is not simple to answer, but very important to the traceability maintenance for high precision CO2 isotope measurements, thus critical for verifying trends in the atmospheric δ13C. Various combinations of Big Delta values amongst different standards and references have provided powerful tools to ensure the QAQC

25

procedures in our program. These are described as follows:

(17)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

– The comparison of δ_NBS1813 _/_VPDB-CO2 between our values and the IAEA rec-ommended values ensures the quality of _∆_NBS18_/_NBS19-CO2, _∆_Cal1_/_NBS19-CO2,

∆Cal2/NBS19-CO2,∆Cal2/Cal1during annual calibrations.

– The comparison of corresponding Big Delta values on ∆45 between two IRMSs (MAT252 and IsoPrime) ensures the consistency of the acid digestion procedures

5

(the comparison in∆46is not valid due to the difference in high vacuum and water levels between the two instruments).

– The largest Big Delta values of ∆Cal2/Cal1 set the benchmark values of the ion source’s cleanliness. An obvious drift away from these values indicates the degra-dation from the benchmark condition.

10

– The comparison in∆Cal2/Cal1between the daily measured values with these from annual calibrations validates the quality of the Cal1s and Cal2s used for daily measurements to ensure routine individual measurements firmly anchored to the primary scale.

– To ensure that the primary and secondary standards have not drifted over time,

15

a batch of uniformly pure CO₂ samples has been made by freezing them into ampoules followed by flame sealing. The variation ranges ofδ13C andδ18O are less than 0.02 ‰ and 0.04 ‰, respectively. This batch of ampoules are only used for annual calibrations, referred to as “annual calibration WRG”. If the pure CO2 preparation procedure is consistent for every year, the rawδ45andδ46 values of

20

the standards (e.g. NBS19, NBS18, Cal1 and Cal2) relative to the “WRG” should be close to constants for each calibration as well as between individual calibra-tions. As shown in Tables 1 to 3, the “annual calibration WRG” ampoules started being used in 2008 and all the corresponding rawδ45 andδ46 values by MAT252 are very consistent from year to year with very small standard deviations (only

25

∼0.01 in δ45 and ∼0.03 in δ46). These results suggests that all the four

(18)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Big Delta values are likely due to the impacts from external factor or instrument condition changes.

6 Challenges and recommendations

A primary anchor on the VPDB scale is required for atmospheric CO₂ isotope mea-surements. Primary anchors can be commercially purchased pure CO2, pure CO2from

5

carbonate or CO2from natural air. The primary anchors should be on the VPDB scale and must be calibrated relative to NBS19-CO₂. All calibrations are based on the as-sumption that NBS19-CO2 has evolved from the NBS19 carbonate properly with the isotopic composition of the assigned values passe on correctly. This assumption is not necessarily valid due to the heterogeneity of NBS19 carbonate, which may vary

10

between di_fferent batches purchased at various times, the uncertainties of pure CO₂ evolved from carbonates (which may be caused by having slightly different reaction temperatures and specific gravities of H3PO4 in acid digestions) and the variations in cleanliness of ion-source and physical configurations, which determines the degree of cross-contamination between samples and WRG during measurements. All of these

15

factors will associate with the uncertainty of primary anchor, then, ultimately impact on the uncertainties of individual isotope measurements through a so-called traceability chain via error propagation. In order to characterize and minimize the uncertainties for isotopic measurements (including both discrete flask and continuous13C measure-ments, e.g. using Cavity-Ring-Down Techniques and other measurement techniques),

20

the following strategy for maintaining traceability has been devised:

– Use different Big Delta values from multiple standards to establish a unique trace-ability pathway, as a documented chain in Eq. (3).

– Clearly identifying the number of levels through the traceability chain (i.e., how many levels of standards are needed to link ambient measurements to the primary

25

(19)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

– Select at least two standards for each level (could be air-CO₂, or pure CO₂evolved from carbonates or commercially pressurized pure CO2) with relatively large iso-topic differences.

– If possible, using the lab standards with the largest Big Delta value, as a QAQC tool, to monitor instrument fluctuations. Using one of the secondary standards as

5

the primary anchor, which should has the δ13C value between NBS19 and the ambient air, and at the same time has a relative small Big Delta value (relative to NBS19-CO2).

– Using the same Working Reference Gas during annual calibration periods for a decadal time span to monitor the fluctuations in raw δ45 and δ46 values for

10

NBS19-CO2and other standards, ensuring the stability of the individual standards by valid acid preparation/extraction procedures. This kind of WRG has to be very homogenous and stable within an uncertainty range of <0.02 ‰ and <0.04 ‰ in δ13C and δ18O, respectively. The pure-CO2 flame sealed in ampoules (e.g. “NARCIS”s type of samples, Mukai et al., 2005) should be ideally used as “annual

15

calibration WRG”.

– Improve the accuracy and precision of the reaction temperature during acid di-gestions of carbonates (e.g. thermometer calibration via a primary device, and temperature and humidity control of the surrounding environment of the whole reaction system).

20

– Use H3PO4within a consistent range of specific gravity (i.e., 1.91–1.92 g cm− 3

).

– If the primary anchor is air CO2, it should be calibrated directly by NBS19-CO2at least once per year.

If these challenges and the recommendations are carefully taken care of, the target of overall uncertainties for individual measurements (i.e., 0.02 ‰ forδ13C and 0.05 ‰ for

25

(20)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Appendix A

Measurement protocols

A1 During annual calibrations

Usually, three sets of pure CO2 ampoules are prepared via acid digestions from car-bonates. Each set include NBS19, NBS18, Cal1 and Cal2. All ampoules are analyzed

5

against the same working reference gas of pure CO2(e.g. APB2, a high pressure cylin-der of pure CO2purchased from Air Products). A calibration event is completed within a time span of one day. The measurements sequence is shown in Table A1.

A2 During routine measurements

Usually, there are total 12 samples measured for a period of one day by the dedicated

10

IRMS (MAT252) together with lab standards (i.e., Cal1 and Cal2). The measurement order and the reasons to carry out this order are shown in the Table A2.

Appendix B

Procedure for preparation of∼100 % phosphoric acid (H3PO4)

B1 Apparatus

15

Hot plate with stirrer option 800 ml pyrex beaker

Teflon coated magnetic stirrer Pyrex spatula

Large metal beaker tongs

(21)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

B2 Chemicals

85 % H3PO4 400 ml (Source: Aldrich cat# 21,510-4) P2O5 300 g (Source: Aldrich cat# 29,822-0) H₂O₂30 % 2 ml (Source: Aldrich cat# 21,676-3)

(H2O2bp: 150.2◦ C, mp:−0.43◦C)

CrO3 10–20 mg (a few flakes) (Source: Aldrich cat#20,782-9)

B3 Procedures

1. Place 85 % phosphoric acid in an 800 ml beaker on the hot plate/with the stirrer in a fume hood and stir on very high speed with magnetic stirrer.

5

2. Very slowly add P2O5.

3. Slowly add H2O2 (to oxidize any possible organic compounds), turn on heat and slowly raise the temperature of the liquid.

4. As it heats, add a few flakes of CrO3 to see if there is any excess H2O2(H2O2is a reducing agent now and can be oxidized by CrO3). The solution may undergo

10

a color change from yellow to light green (Cr6+ to Cr3+).

5. Heat the acid to boiling and allow to boil for 2.5 h.

6. Cool slightly and transfer to Teflon storage bottle, while still hot so that it is still viscous, using the large tongs to hold the beaker.

7. Determine the specific Gravity of the acid once totally cooled down to room

tem-15

(22)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

8. Specific gravity can be measured by pipetting 10 ml of room temperature acid into a volumetric flask that has been pre-weighed on a good quality balance (5 digits). Be careful not to get any acid on the walls of the flask above the volumetric line. Also ensure that the acid is homogenized by shaking the container before pipetting. Stratification of the acid may occur.

5

9. It is better to keep the phosphoric acid (∼100 %) in the Teflon storage bottle for

about two month before using it (according to our experience).

B4 Recommendations for the preparation

1. Transfer the pure H3PO4to the storage bottle while still hot so that it is still viscous. Use rubber gloves, safety glasses and tongs. This step is dangerous.

10

2. Make sure that no water (also no vapor) comes into contact with the H3PO4either during the preparation or after transferring to the Teflon storage bottles to ensure the quality ofδ18O values.

Appendix C

Uncertainty estimation of the traceability in CO2isotope measurements

15

Given:

X= ∆Sam/Lab-Std

Y = ∆Lab-Std/NBS19-CO2

A= ∆NBS19-CO2/VPDB-CO2=constant

20

(23)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

f(X,Y)₌δ45/46(CO₂)_Sam_/_VPDB-CO2(X,Y)

=[(X×10−3+1)×(Y ×10−3+1)×(A×10−3+1)−1]×103

=[A×X×Y ×10−6+A×X×10−3+A×Y ×10−3+X×Y ×10−3+A +X+Y]

≈[A×X×10−3+A×Y ×10−3+X×Y ×10−3+A +X+Y] (C1)

5

GivenσX andσY,σf can be calculated as:

σf =[(∂(A×X×10−3)/∂X)2×(σX)2+(∂(A×Y ×10−3)/∂Y)2×(σY)2

+(∂(X×Y ×10−3)/∂X)2×(σX)2+(∂(X×Y×10−3)/∂Y)2×(σY)2+(σX)2

+(σY)2]1/2=[A2×10−6×(σX)2+A2×10−6×(σY)2+Y2×10−6×(σX)2

10

+X2

×10−6×(σY)2+(σX)2+(σY)2]1/2

≈[(σX)2+(σY)2]1/2 (C2)

Acknowledgement. This work has been supported by Environment Canada (EC) A-base fund-ing (related to Climate Research) over the last 10 yr. The authors sincerely thank Ann-Lise Norman for her effort during the early stage of this work (before 1999); those research

man-15

agers from Environment Canada for their understanding and supporting the fundamental work in carbon cycle related atmospheric measurements; Colleagues from Global Monitoring Di-vision (GMD) in Earth System Research Laboratory (ESRL), National Oceanic Atmospheric Administration (NOAA), US and The Institute of Arctic and Alpine Research (INSTAAR) at Uni-versity of Colorado for the accessibility of the global δ13C (CO₂) data; World Meteorological

20

(24)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

References

Allison, C. E. and Francey, R. J.: Verifying Southern Hemisphere trends in atmo-spheric carbon dioxide stable isotopes, J. Geophys. Res.-Atmos., 112, D21304, doi:10.1029/2006JD007345, 2007.

Allison, C. E., Francey, R. J., and Meijer, H. A. J.: Recommendations for reporting stable isotope

5

measurements of carbon and oxygen in CO₂gas, IAEA TECDOC, 825, 155–162, 1995. Bakwin, P. S., Tans, P. P., White, J. W. C., and Andres, R. J.: Determination of the isotopic

(13C/12C) discrimination by terrestrial biology from a global network of observations, Global Biogeochem. Cy., 12, 555–562, 1998.

Battle, M., Bender, M. L., Tans, P. P., White, J. W. C., Ellis, J. T., Conway, T., and Francey, R. J.:

10

Global carbon sinks and their variability inferred from atmospheric O₂ andδ13C, Science, 287, 2467–2470, 2000.

Brand, W. A., Huang, L., Mukai, H., Chivulescu, A., Richter, J. M., and Rothe, M.: How well do we know VPDB? Variability ofδ13C andδ18O in CO₂generated from NBS19-calcite, Rapid Commun. Mass Sp., 23, 915–926, 2009.

15

Ciais, P., Tans, P. P., Trolier, M., White, J. W. C., and Francey, R. J.: A large Northern Hemisphere terrestrial CO₂sink indicated by the13C/12C Ratio of atmospheric CO₂, Science, 269, 1098– 1102, 1995.

Ciais, P., Scott, A., Denning, Tans, P. P., Berry, J. A., Randall, D. A., Collatz, G. J., Sellers, P. J., White, J. W. C., Trolier, M., Meijer, H. A. J., Francey, R. J., Monfray, P., and Heimann, M.:

20

A three-dimensional synthesis study ofδ18O in atmospheric CO₂. 1. Surface fluxes, J. Geo-phys. Res., 102, 5857–5872, 1997.

Clayton, R. N.: Oxygen isotope fraction in the system calcium carbonate-water, J. Chem. Phys., 30, 1246–1250, 1959.

Coplen, T. B., Brand, W. A., Gehre, M., Groning, M., and Meijer, H. A. J.: After two decades

25

a second anchor for the VPDBδ13C scale, Rapid Commun. Mass Sp., 20, 3165–3166, 2006. Craig, H.: Isotopic standards for carbon and oxygen and correction factors for

mass-spectrometric analysis of carbon dioxide, Geochim. Cosmochim. Ac., 12, 133–149, 1957. Francey, R. J., Tans, P. P., Allison, C. E., Enting, I. G., White, J. W. C., and Trolier, M.: Changes

in oceanic and terrestrial carbon uptake since 1982, Nature, 373, 326–330, 1995.

30

Flesch, G. D., Anderson Jr., A. R., and Svec, H. J.: A secondary isotopic standard for6Li/7Li determinations, Int. J. Mass Spectrom., 12, 265–272, 1973.

(25)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

GAW Report No. 161: The Expert Group Recommendation, in: Report of the 12th WMO/IAEA Meeting of Experts on carbon dioxide and related tracers measurement techniques, edited by: Worthy, D. and Huang, L., Toronto, Canada, 15–18 September 2003, 2005.

GAW Report No. 168: The Expert Group Recommendation, in: Report of the 13th WMO/IAEA Meeting of Experts on carbon dioxide and related tracers measurement techniques, edited

5

by: Miller, J., Boulder, USA, 19–22 September 2005, 2006.

GAW Report No. 186: The Expert Group Recommendation, in: Report of the 14th WMO/IAEA Meeting of Experts on carbon dioxide, other Greenhouse Gases and related tracers mea-surement techniques, edited by: Laurila, T., Helsinki, Finland, 10–13 September 2007, 2009. GAW Report No. 194: the Expert Group Recommendation, in: Report of the 15th WMO/IAEA

10

Meeting of Experts on carbon dioxide, other greenhouse gases and related tracers measure-ment techniques, edited by: Brand, W. A., Jena, Germany, 7–10 September 2009.

Gonfiantini, R., Stichler, W., and Rozanski, K.: Standards and Intercomparison Materials Dis-tributed by the International Atomic Energy Agency for Stable Isotope Measurements, Pro-ceedings of a consultants meeting held in Vienna, 1–3 December 1993, IAEA-TECDOC-825,

15

13–29, 1995.

Huang, L. and Worthy, D.: Flask Sampling Strategy by MSC for Atmospheric Observations of Greenhouse Gases and CO₂Isotopes over Canada: Status and Plan, in: Report of the 12th WMO/IAEA Meeting of Experts on CO₂ Concentration and Related Tracer Measurement Techniques, Toronto, September 2003, GAW publication, 161, 94–100, 2005.

20

Huang, L., Norman, A. L., Allison, C. E., Francey, R. J., Ernst, D., Chivulescu, A., and Higuchi, K.: Traceability – maintenance for high precision stable isotope measurement (δ13C andδ18O) of Air-CO₂ by Lab-Carbonate-Standards at MSC: Application to the Inter-Comparision Program (alert, Canada) with CSIRO, in: Report of the 11th WMO/IAEA Meet-ing of Experts on CO₂ Concentration and Related Tracer Measurement Techniques, Tokyo,

25

September 2001, WMO publication, No. 148, 9–16, 2002.

Keeling, C. D.: The concentration and isotopic abundances of carbon dioxide in the atmosphere, Tellus, 12, 200–203, 1960.

Keeling, C. D.: The concentration and isotopic abundance of carbon dioxide in rural and marine air, Geochim. Cosmochim. Ac., 24, 277–208, 1961.

30

(26)

AMTD

5, 4003–4040, 2012

L. Huang et al.

Title Page

Tables Figures

◭ ◮

Back Close

Full Screen / Esc

Discussion

P

a

per

|

Dis

cussion

P

a

per

|

Discussion

P

a

per

|

Discussio

n

P

a

per

|

Keeling, C. D., Whorf, T. P., Wahlen, M., and van der Plicht, J.: Interannual extremes in the rate of rise of atmospheric carbon dioxide since 1980, Nature, 375, 666–670, 1995.

Masarie, K. A. and Tans, P. P.: Extension and integration of atmospheric carbon dioxide data into a globally consistent measurement record, J. Geophys. Res.-Atmos., 100, 11593–11610, 1995.

5

Meijer, H. A. J., Neubert, R. E. M., and Visser, G. H.: Cross contamination in dual inlet isotope ratio mass spectrometers, Int. J. Mass Spectrom., 198, 45–61, 2000.

McCrea, J. M.: On the isotopic chemistry of carbonates and a paletemperature scale, J. Chem. Phys., 18, 849–857, 1950.

Mook, W. G., Koopmans, M., Carter, A. F., and Keeling, C. D.: Seasonal, latitudinal, and secular

10

variations in the abundance and isotopic ratios of atmospheric carbon dioxide. 1. Results from land stations, J. Geophys. Res., 88, 10915–10933, 1983.

Mukai, H., Nakazawa, T., Brand, W. A., Huang, L., Levin, I., Allison, C., White, J., Leuen-berger, M., and Assonov, S. S.: About disagreements in inter-comparison activities of iso-tope ratio measurements for CO₂, in: Report of the 13th IAEA/WMO Meeting of CO₂experts,

15

edited by: Miller, J. and Conway, T., Vol. WMO/GAW, No. 168, Boulder, 41–48, 2005. Sharp, Z.: The phosphoric acid method, Chapt. 6 in: Principles of Stable Isotope Geochemistry,

Pearson, Prentice Hall, 121–122, 2007.

Stichler, W.: Interlaboratory comparison of new materials for carbon and oxygen isotope ratio measurements, Proceedings of a Consultants Meeting held in Vienna, 1–3 December 1993,

20

IAEA-TECDOC-825, 67–74, 1995.

Verkouteren, M. R. and Klinedinst, D. B.: Value assignment and uncertainty estimation of se-lected light stable isotope reference materials: RMs 8543–8545, RMs 8562–8564, and RM 8566, in NIST Special Publication 260-149, 2004.

Wendeberg, M., Richter, J. M., and Rothe, M., and Brand, W.:δ18O anchoring to VPDB: calcite

25

digestion with 18O adjusted ortho-phosphoric acid, Rapid Commun. Mass Sp., 25, 851–860, doi:10.1002/rcm.4933, 2011.

White, J. W. C. and Vaughn, B. H.: University of Colorado, Institute of Arctic and Alpine Re-search (INSTAAR), Stable Isotopic Composition of Atmospheric Carbon Dioxide (13C and

18

O) from the NOAA ESRL Carbon Cycle Cooperative Global Air Sampling Network, 1990–

30

2010, Version: 2011–11-08, available at: ftp://ftp.cmdl.noaa.gov/ccg/co2c13/flask/event/ (last access: 30 May 2012), 2011.

Maintaining consistent traceability in high precision isotope measurements of CO<sub>2</sub>: verifying atmospheric trends of &delta;<sup>13</sup>C

AMTD

Maintaining consistent traceability in high

precision isotope measurements of CO

2

:

verifying atmospheric trends of

δ

13

C

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

AMTD

Maintaining consistent traceability in high precision isotope measurements of CO<sub>2</sub>: verifying atmospheric trends of δ<sup>13</sup>C

₂

_C